# Feature Flag Software: Pick One, Then Ship Behind It URL: https://whatshouldibuildnext.com/lp/feature-flag-software Type: landing Locale: en Published: 2026-10-25 Updated: 2026-10-28 --- > LaunchDarkly, PostHog, GrowthBook and Unleash compared: how to choose feature flag software, when a plain boolean is enough, and what to flag first. *For devs shipping alone* ## Feature Flag Software, Picked Without the Hype Compare feature flag software by what a solo dev or small team actually needs, then scope a project worth putting behind a flag. Free, no signup. ## Four jobs a feature flag does ### Ship dark, release later Merge unfinished work to main behind a flag. Deploying code and releasing a feature stop being the same event, which makes Friday merges less scary. ### Roll out gradually Enable a feature for yourself, then 5% of users, then everyone. If error rates climb, you stop at 5% instead of apologising to all of them. ### Kill switch in seconds A flaky third-party integration or a slow query can be switched off from a dashboard. No rollback, no redeploy, no 3 am hotfix branch. ### Test with real users Serve two variants, measure one metric, keep the winner. Some tools bundle experimentation, others leave it to your analytics stack. ## Four tools, honest differences | Criteria | LaunchDarkly | PostHog | GrowthBook | Unleash | |---|---|---|---|---| | License | Proprietary | MIT | MIT | Apache | | Self-host option | No | Yes | Yes | Yes | | Free tier | Yes, capped on usage | Yes, 1M flag requests a month | Yes on cloud, free to self-host | Free to self-host | | Built-in experiments | Yes | Yes, in cloud | Yes | Limited | | Best fit | Teams that want a managed enterprise product | Teams that also want product analytics | Teams that care about A/B testing | Teams that need data sovereignty | *Use case* ## Don't adopt a tool before you have a risky feature The usual mistake is installing feature flag software on a project that has nothing worth flagging. A flag earns its keep when there is a change you would hesitate to ship to everyone: a payment flow, a data migration, an AI feature with unpredictable cost. If you can't name that change, a boolean in an environment variable is enough for now. The generator on whatshouldibuildnext.com turns a rough idea into a spec with a suggested stack and the one or two features that carry real risk, so you know what to put behind a flag before you pick a vendor. - Free, client-side, no signup - Names the features that are risky to ship - Suggests a stack that supports your flag SDK *Hosting* ## Hosted or self-hosted: pick the smaller burden Open-source feature flag tools are free in licence terms, not in time. Self-hosting means a database to back up, a service to keep alive and an upgrade path to follow, all for a system your production code now depends on at request time. Hosted tools remove that, and their free tiers are generous enough for most side projects. Choose self-hosting when data residency or per-request cost forces your hand, not because it feels more indie. Whichever you pick, code against OpenFeature so the choice stays reversible. - Hosted: no servers, usage-based pricing - Self-hosted: no request bill, you own uptime - OpenFeature keeps the vendor swappable ## From zero to a flagged feature 1. **Name the risky change** — Pick the one feature that could hurt you if it shipped broken. That is your first flag, not all twenty. 2. **Start with the free tier** — Create an account or spin up the open-source image locally, grab an SDK key and wire up one flag in a branch. 3. **Default to off** — Make the safe behaviour the fallback so a failed flag lookup never exposes unfinished code. 4. **Roll out in stages** — Yourself first, then a small percentage, then everyone. Watch errors and latency between each step. 5. **Delete the flag** — Once a feature sits at 100% and is stable, remove the flag and the dead code path in one pull request. ## Common questions ### What is feature flag software? It is a tool that lets you turn code paths on or off at runtime without redeploying. You wrap a risky change in a flag, ship it dark, then enable it for yourself, a percentage of users or everyone, and switch it off instantly if it breaks. ### Do I need feature flag software on a solo side project? Usually not on day one. A boolean in an environment variable covers a lot. Reach for a real tool when you need per-user targeting, gradual rollouts or a kill switch you can flip without touching a server. ### Is there a free feature flag tool? Yes. LaunchDarkly, PostHog, GrowthBook, Flagsmith and Statsig all list free tiers, and Unleash, GrowthBook, Flagsmith and PostHog have open-source editions you can self-host. Limits and prices change, so check each pricing page before you commit. ### Should I self-host or use a hosted service? Hosted means no servers to patch and flags evaluated for you. Self-hosted means no per-request bill and full data control, but you run the database and the uptime. For a side project, hosted free tiers are usually less work. ### What is OpenFeature and does it matter? OpenFeature is a vendor-neutral API for flag evaluation. If you code against it, switching providers later means changing a provider plugin, not rewriting every flag call. It is cheap insurance against lock-in. ### How do I stop old flags from piling up? Give every flag an owner and a removal date when you create it. Treat a flag that has been at 100% for a few weeks as debt: delete the flag and the dead branch of code in the same pull request. ### What should I build to practice with flags? Something with at least one feature you would be nervous to release to everyone. The generator on whatshouldibuildnext.com scopes a small project with a risky feature or two, which gives you a real reason to wrap code in a flag. ## Scope something worth flagging Free spec generator, client-side, no signup. Get the features and a suggested stack, then pick your flag tool. *Call to action: Generate a project spec* ## FAQ ### What is feature flag software? It is a tool that lets you turn code paths on or off at runtime without redeploying. You wrap a risky change in a flag, ship it dark, then enable it for yourself, a percentage of users or everyone, and switch it off instantly if it breaks. ### Do I need feature flag software on a solo side project? Usually not on day one. A boolean in an environment variable covers a lot. Reach for a real tool when you need per-user targeting, gradual rollouts or a kill switch you can flip without touching a server. ### Is there a free feature flag tool? Yes. LaunchDarkly, PostHog, GrowthBook, Flagsmith and Statsig all list free tiers, and Unleash, GrowthBook, Flagsmith and PostHog have open-source editions you can self-host. Limits and prices change, so check each pricing page before you commit. ### Should I self-host or use a hosted service? Hosted means no servers to patch and flags evaluated for you. Self-hosted means no per-request bill and full data control, but you run the database and the uptime. For a side project, hosted free tiers are usually less work. ### What is OpenFeature and does it matter? OpenFeature is a vendor-neutral API for flag evaluation. If you code against it, switching providers later means changing a provider plugin, not rewriting every flag call. It is cheap insurance against lock-in. ### How do I stop old flags from piling up? Give every flag an owner and a removal date when you create it. Treat a flag that has been at 100% for a few weeks as debt: delete the flag and the dead branch of code in the same pull request. ### What should I build to practice with flags? Something with at least one feature you would be nervous to release to everyone. The generator on whatshouldibuildnext.com scopes a small project with a risky feature or two, which gives you a real reason to wrap code in a flag.